Research On Trace Transform And Its Geophysical Applications In Seismic Surface Wave And Random Noise Attenuation

Seismic exploration is an important tool in development of oil-gas and mineral resources. It is so important for the natural resource exploration that we should attenuate the noise and raise the signal to noise ratio (SNR) of the seismic records as much as possible. The seismic noise could be divided into two types concerning with the characteristics on the seismic profile: regular noise and irregular noise. The regular noise, also called the coherent noise, mainly contains: surface waves, sound waves and multiples, etc. The irregular noise is the so called random noise. In this paper, we have a systematic and detailed research on the Trace transform from image processing and import this theory into seismic record denoising. The proposed Co-Core Trace transform filtering and radial Trace Time-frequency peak filtering aims at seismic surface wave attenuation and seismic random noise attenuation, respectively.Based on the Trace transform, this paper first uses a common-shot record to discuss the construction of the specific Trace function in the seismic surface wave processing. Combining with the spatial and temporal propagation, energy distribution and other characteristics of the surface waves, we construct a Co-Core Trace function to differentiate the surface wave energy from the reflected wave energy in the Trace transform domain. Then analyze the range of the Co-Core weight coefficient and energy distributions of different types of seismic waves. Besides, we propose the concept of the relative threshold to separate the surface wave off from the transform domain. All of the above analysas make the Co-Core Trace transform filtering feasible for the surface wave attenuation of the seismic records. Due to the conspicuous disparity, the energies of the surface waves are significantly enhanced and reflected energy remains at a relatively low level in the Co-Core Trace transform domain. Furthermore, the enhanced energy is always two orders of magnitude greater than the refleceted energy. Therefore, with the relative threshold, we could keep the reflected signal intact and meanwhile cleanly wipe off the surface waves in one time. In this paper, we also discuss the sampling interval of the Trace lines and the relative threshold through experiments on synthetice seismic records. Empirical optimal values are given with respect to both the quality and the computation time. Experiments with synthetic records and field data demonstrate that with the appropriate parameters, the proposed Co-Core Trace transform filtering performs better both in terms of surface wave attenuation and reflected signal preservation than some conventional methods, such as theτ? pmethod, F-K transform filtering and wavelet filtering. Besides, the Co-Core Trace transform filtering can attenuate some D.C. seismic noise. This paper also discusses the texture of the seismic record with the triple feature from the image texture recognition point of view. After transforming a seismic record into the Trace transform domain with the distanceρand the angleΦ, we could go on calculate functions aboutρ, then aboutΦto attain a triple feature of the original seismic record. The triple features could reflect the texture features of a seismic record. In this paper, we give the definitions of invariant functional and sensitive functional, and then discuss in detail several combinations of the above two functionals to dereive the conditions of invariant triple feature. We also point that different triple features reflect different texture features of a seismic record. To a given functional combination, similar seismic records could attain similar triple features. We prove with several images and also utilize this property to test the difference between the ideal result and different filted results of the seismic records. Experiments with several triple features illustrate that the Co-Core Trace transform filtered record not only be effective in surface wave attenuation, but also recover the texture to the ideal level, and the texture features could reflect the statistical properties of the seismic record as well as the peak values, dominant frequency, and signal to noise ratio, etc.In order to deal with the random noise in the seismic records, this paper proposes a radial Trace time-frequency peak filtering (TFPF) algorithm. Taking the application conditions, error resources and limitations of the conventional TFPF fully into account, the proposed method rotates the seismic reflection events in the radial Trace transform domain, strech the apparent time and raise the linearity of the transformed seismic wavelet, which is also based on the correlation between the adjacent channels. Applying the TFPF in the radial Trace transform domain could reduce the error at the source, and problems such as amplitude degradation, width change and wave distortion could be solved. The radial Trace TFPF can better recover the reflected signal and raise the SNR of the seismic records to a higher level. Experiments on synthetic models show that the radial Trace TFPF can provide better performance in both random noise attenuation and reflected signal preservation with a fix window length than the conventional TFPF, and the result is no longer much influenced by the window length as the conventional TFPF. Experiments on field data prove that the reflection events become continuous and compact with little energy loss after the radial Trace TFPF, and some originally buried events are also revealed. Besides, this proposed algorithm provides basis for time-space TFPF technical in seismic random noise attenuatin with other types of Trace combinations.